♻️ Refactor

lamindb/core/_run_context.py

@@ -11,7 +11,6 @@
 from lamin_utils import logger
 from lamindb_setup import settings as setup_settings
 from lamindb_setup.core import InstanceSettings
-from lamindb_setup.core.hashing import hash_code
 from lamindb_setup.core.types import UPathStr
 from lnschema_core import Run, Transform, ids
 from lnschema_core.types import TransformType
@@ -20,6 +19,7 @@
 from lamindb.core._transform_settings import transform_settings
 
 from ._settings import settings
+from ._sync_git import get_transform_reference_from_git_repo
 
 is_run_from_ipython = getattr(builtins, "__IPYTHON__", False)
 
@@ -204,76 +204,6 @@ def raise_transform_settings_error() -> None:
     )
 
 
-def clone_git_repo(git_url: str) -> None:
-    if not git_url.endswith(".git"):
-        git_url += ".git"
-    logger.important(f"cloning {git_url}")
-    result = subprocess.run(
-        f"git clone --depth 10 {git_url}",
-        shell=True,
-        capture_output=True,
-    )
-    if result.returncode != 0:
-        raise RuntimeError(result.stderr.decode())
-
-
-def dir_from_repo_url(repo_url: Optional[str]) -> Optional[str]:
-    if repo_url is not None:
-        cd_repo = repo_url.split("/")[-1].replace(".git", "")
-    return cd_repo
-
-
-def get_git_commit_hash(
-    blob_hash: str, cd_repo: Optional[str]
-) -> subprocess.CompletedProcess:
-    return subprocess.run(
-        f"git log --find-object={blob_hash} --pretty=format:%H",
-        shell=True,
-        capture_output=True,
-        cwd=cd_repo,
-    )
-
-
-def get_filepath_within_git_repo(
-    commit_hash: str, blob_hash: str, cd_repo: Optional[str]
-) -> str:
-    result = subprocess.run(
-        f"git ls-tree -r {commit_hash} | grep -E {blob_hash}",
-        shell=True,
-        capture_output=True,
-        cwd=cd_repo,
-    )
-    if result.returncode != 0:
-        raise RuntimeError(
-            f"git ls-tree -r {commit_hash} | grep -E {blob_hash}\n"
-            + result.stderr.decode()
-        )
-    if len(result.stdout.decode()) == 0:
-        raise RuntimeError("Could not find filepath within git repo.")
-    filepath = result.stdout.decode().split()[-1]
-    return filepath
-
-
-def get_transform_reference_from_git_repo(path: Path):
-    blob_hash = hash_code(path).hexdigest()
-    cd_repo = None
-    result = get_git_commit_hash(blob_hash, cd_repo=None)
-    commit_hash = result.stdout.decode()
-    if commit_hash == "" or result.returncode == 1:
-        cd_repo = dir_from_repo_url(settings.sync_git_repo)
-        clone_git_repo(settings.sync_git_repo)
-        result = get_git_commit_hash(blob_hash, cd_repo=cd_repo)
-        commit_hash = result.stdout.decode()
-        if commit_hash == "" or result.returncode == 1:
-            raise RuntimeError(
-                f"Did not find file in git repo\n{result.stderr.decode()}"
-            )
-    gitpath = get_filepath_within_git_repo(commit_hash, blob_hash, cd_repo)
-    reference = f"{settings.sync_git_repo}/blob/{commit_hash}/{gitpath}"
-    reference_type = "url"
-    return reference, reference_type
-
-
 class run_context:
     """Global run context."""
 

lamindb/core/_sync_git.py

@@ -0,0 +1,78 @@
+import subprocess
+from pathlib import Path
+from typing import Optional
+
+from lamin_utils import logger
+from lamindb_setup.core.hashing import hash_code
+
+from ._settings import settings
+
+
+def clone_git_repo(git_url: str) -> None:
+    if not git_url.endswith(".git"):
+        git_url += ".git"
+    logger.important(f"cloning {git_url}")
+    result = subprocess.run(
+        f"git clone --depth 10 {git_url}",
+        shell=True,
+        capture_output=True,
+    )
+    if result.returncode != 0:
+        raise RuntimeError(result.stderr.decode())
+
+
+def dir_from_repo_url(repo_url: Optional[str]) -> Optional[str]:
+    if repo_url is not None:
+        cd_repo = repo_url.split("/")[-1].replace(".git", "")
+    return cd_repo
+
+
+def get_git_commit_hash(
+    blob_hash: str, cd_repo: Optional[str]
+) -> subprocess.CompletedProcess:
+    return subprocess.run(
+        f"git log --find-object={blob_hash} --pretty=format:%H",
+        shell=True,
+        capture_output=True,
+        cwd=cd_repo,
+    )
+
+
+def get_filepath_within_git_repo(
+    commit_hash: str, blob_hash: str, cd_repo: Optional[str]
+) -> str:
+    result = subprocess.run(
+        f"git ls-tree -r {commit_hash} | grep -E {blob_hash}",
+        shell=True,
+        capture_output=True,
+        cwd=cd_repo,
+    )
+    if result.returncode != 0:
+        raise RuntimeError(
+            f"git ls-tree -r {commit_hash} | grep -E {blob_hash}\n"
+            + result.stderr.decode()
+        )
+    if len(result.stdout.decode()) == 0:
+        raise RuntimeError("Could not find filepath within git repo.")
+    filepath = result.stdout.decode().split()[-1]
+    return filepath
+
+
+def get_transform_reference_from_git_repo(path: Path):
+    blob_hash = hash_code(path).hexdigest()
+    cd_repo = None
+    result = get_git_commit_hash(blob_hash, cd_repo=None)
+    commit_hash = result.stdout.decode()
+    if commit_hash == "" or result.returncode == 1:
+        cd_repo = dir_from_repo_url(settings.sync_git_repo)
+        clone_git_repo(settings.sync_git_repo)
+        result = get_git_commit_hash(blob_hash, cd_repo=cd_repo)
+        commit_hash = result.stdout.decode()
+        if commit_hash == "" or result.returncode == 1:
+            raise RuntimeError(
+                f"Did not find file in git repo\n{result.stderr.decode()}"
+            )
+    gitpath = get_filepath_within_git_repo(commit_hash, blob_hash, cd_repo)
+    reference = f"{settings.sync_git_repo}/blob/{commit_hash}/{gitpath}"
+    reference_type = "url"
+    return reference, reference_type